FIW Working Paper N 46 February How to Measure Globalisation? A New Globalisation Index (NGI) Petra Vujakovic. Abstract

Transcription

1 FIW Working Paper FIW Working Paper N 46 February 2010 How to Measure Globalisation? A New Globalisation Index (NGI) Petra Vujakovic Abstract In this article, a new composite globalisation index will be presented. With its 21 variables, it accounts for the multidimensionality of this phenomenon instead of relying purely on economic indicators. As compared to other existing globalisation indices, three major innovations are introduced in this New Globalisation Index (NGI). Firstly, five variables that have until now not been used in globalisation indices enter the calculations. Secondly, geographical distances between countries are incorporated into the index in the trade variable, so as to account for the distinction between globalisation and regional integration. A final innovation is a methodological one, which concerns the use of a statistical method (principal component analysis) to form subindices according to the statistical features of the variable structure. A control for country size is employed for significantly affected variables, as was done in some other globalisation indices before. The final index contains 70 countries and covers a period between 1995 and JEL Codes: F15, F20, F36, G15, P51 Keywords: globalisation; openness of economies; measurement of openness; measurement of globalisation; integration; composite index The author Joint Vienna Institute, petravujakovic@gmail.com FIW, a collaboration of WIFO ( wiiw ( and WSR (

2

3 How to Measure Globalisation? A New Globalisation Index (NGI) Petra Vujakovic Abstract: In this article, a new composite globalisation index will be presented. With its 21 variables, it accounts for the multidimensionality of this phenomenon instead of relying purely on economic indicators. As compared to other existing globalisation indices, three major innovations are introduced in this New Globalisation Index (NGI). Firstly, five variables that have until now not been used in globalisation indices enter the calculations. Secondly, geographical distances between countries are incorporated into the index in the trade variable, so as to account for the distinction between globalisation and regional integration. A final innovation is a methodological one, which concerns the use of a statistical method (principal component analysis) to form subindices according to the statistical features of the variable structure. A control for country size is employed for significantly affected variables, as was done in some other globalisation indices before. The final index contains 70 countries and covers a period between 1995 and 2005.

4 CONTENTS 1. Introduction Definition of Globalisation The Choice of Indicators for the New Globalisation Index The Effect of Geographical Distance Other Adjustments of the Data Construction of Factors and Weights Main Results Changes over Time Comparing with Results of Other Globalisation Indices Conclusions and Further Issues References FIGURES Figure 1: Contribution of individual dimensions to total globalisation score for selected countries Figure 2: Globalisation over time Bulgaria, Romania, Croatia and Turkey TABLES Table 1: List of variables in the New Globalisation Index... 6 Table 2: Difference in rankings for 'trade in goods' variable with and without distanceweights Table 3: Regression results of the adjustment for country size Table 4: Dimensions and Weights Table 5: New Globalisation Index 2005 rankings and scores Table 6: Top 10 (rankings, 2005) Table 7: Big developed countries (rankings, 2005) Table 8: "New Europe" (rankings, 2005) Table 9: BRICSAM countries (rankings, 2005) Table 10: Comparison of the NGI and KOF Index Rankings (R)

5 1. Introduction There are not many expressions as controversial as globalisation, as it combines many contradictory issues under a single concept. It can be good and bad at the same time. It stands for growing power of multinationals in every corner of the world as well as for the worldwide spreading of knowledge and human rights. With the many faces it has, it affects every sphere of life wealth, freedom, cultural habits, health All around the world, it changes tastes of food, influences art, puts new products on market shelves. In this way, it has become a key topic in many discussions, both in coffeehouses and in university institutes. The goal of this paper is not to add more fire to the controversial discussion on the consequences of globalisation. Instead, its purpose is to measure the extent and the relative speed of globalisation, in order to provide some insights about the process of globalisation itself in quantitative terms. Quantifying globalisation is a tricky task, taking into account its complexity and multidimensionality. It is nevertheless worth a try, as such a measure could contribute greatly to the whole globalisation debate. Not only would an index of globalization deepen the understanding of the concept and give an impression of its extent and relative position of countries, but also enable further research of the links between globalisation and other phenomena such as poverty, development, economic growth, etc. The construction of the New Globalisation Index (NGI) in this paper builds on existing literature on globalisation indices and presents a couple of new suggestions for the methodology. One of them is the introduction of five new variables that have until now not been used in other published globalisation indices. The new variables make use of existing 3

6 information relevant for globalisation to add some new aspects to the measure. Secondly, a parameter that accounts for geographical distances between countries is introduced to the index, so as to address the problem of distinction between globalisation and regionalisation. Traditional measures of globalisation are mostly measures of openness, such as the ratio of international trade to GDP, which do not distinguish between interaction with distant countries and relations to neighbouring countries. Regional integration should, however, not be mistaken for globalisation. In order to distinguish between these two phenomena, geographical distances between interacting parties have to be accounted for. A suggestion on how to do this in a globalisation index will be showcased on the example of international trade in goods. A final innovation involves the construction of dimensions of the index according to the pattern structure of the variable set based on the statistical characteristic of the data. A statistical method (principal component analysis) will be employed to identify dimensions according to which the variables will be grouped into subindices of the overall globalisation index. Coupled with the theoretical basis that the variable selection is based on, this step will produce dimensions that are both statistically plausible and conceptually relevant. 2. Definition of Globalisation A desirable starting point for any measurement is a clear idea of the concept that is to be measured. In a world where everybody seems to have an opinion on globalisation, it may come as a surprise that one clear and commonly accepted definition for it does not exist. While some would say the economic globalisation is the motor behind the whole process, others may counter them by naming the cultural or political aspect as the most important one. Caselli (2006) mentions three dimensions of globalisation mainly accepted in contemporary 4

7 theory, namely economic, political and cultural, which may be further divided into subdimensions. A definition of globalisation should reflect this multidimensionality. Also, the fact that globalisation really does enfold the whole globe needs to be stressed, as this is the distinguishing factor between globalisation and other forms of international openness of countries. Exactly the intensity of contact in so many spheres with such intensity at such large distances is the main feature of what we call globalisation today. Owing to it, policymakers, investors and consumers can concentrate finding certain desired products, information or people, without being constrained by their remoteness. This means that the decision between going to the other side of the world and to a neighbouring country may with time and globalisation progress depend increasingly on features other than distance, as the latter becomes easier surmountable and therefore less relevant. Keeping in mind the relevant points from this short discussion, the following definition seems to me suitable at this stage: Globalisation is a process of growing interaction and interdependence between economies, societies and nations across large distances. 3. The Choice of Indicators for the New Globalisation Index Composite indicators usually measure multidimensional concepts which cannot be captured by a single indicator (JRC/OECD, 2008, p. 13). Thus, as the expression implies, a composite indicator is composed of more individual indicators, each capturing a part of the picture. A proper set of indicators is fundamental for the relevance of the final composite index. For the NGI, the search for indicators resulted in a set of 21 variables, presented in the Table 1. 5

8 SOCIAL POLITICAL ECONOMIC Table 1: List of variables in the New Globalisation Index Indices and Variables Source Definition Trade in Goods (weighted with geographical distances) in %GDP UN COMTRADE CEPII Bilateral imports and exports of goods. Data in % GDP. Geographical distances between countries in km, using city-level data to assess the geographic distribution of population inside each country. Trade in Services WB 08 Sum of services exports and imports. Data in % GDP. FDI Stock UNCTAD Sum of inward and outward foreign direct investment stock. Data in % GDP. FDI Flow UNCTAD Sum of inflows and outflows of foreign direct investment recorded in the balance of payments financial account. Data in % GDP. Portfolio Investment Stock (NEW) Portfolio Investment Flow Income Payments to Foreign Nationals Trademark Applications by Non- Residents (NEW) Patent Applications by Non-Residents (NEW) Environmental Agreements (NEW) International Organization Membership IMF 08 IMF 08 WB 08 WB 08 Sum of portfolio investment stock assets and liabilities from the international investment position records. Data in % GDP. Sum of inflows and outflows of portfolio investment recorded in the balance of payments. Data in % GDP. Sum of receipts and payments of employee compensation for non-resident workers, and investment income. Data in % GDP. Share of applications by non-residents to register a trademark with a national or regional trademark office. Data provided by the WIPO. WB 08 Share of patent applications filed by non-residents with a national patent office. Data provided by the WIPO. CIA Absolute number of international environmental agreements ratified. CIA Absolute number of memberships in international organizations. Embassies in Country EWY Absolute number of embassies in a country. Participation in UN Peacekeeping Missions UNDPKO Peacekeeping personnel contributions to UN peacekeeping missions. Migration Stock WB 08 Number of people born in a country other than that in which they live. It includes refugees. Data in percent of total population. International Tourism 1 WB 08 Sum of arrivals and departures of international tourists as a share of population. Outbound Student Mobility (NEW) UNESCO The number of students from a given country studying abroad as a percentage of the total tertiary enrolment in that country. International Phone Calls WB 08 Sum of international incoming and outgoing telephone International Internet Bandwidth WB 08 traffic (in minutes) divided by total population. Contracted capacity of international connections between countries for transmitting Internet traffic. Data in bits per person. International Trade in Newspapers UN COMTRADE Sum of exports and imports in newspapers and periodicals, code of SITC. Data in % GDP. International Trade in Books UN COMTRADE Sum of exports and imports in books and pamphlets, code of the SITC. Data in % GDP. Transfers WB 08 Sum of current transfers recorded in the balance of payments whenever an economy provides or receives goods, services, income, or financial items without a quid pro quo. Data in % GDP. 1 The international tourism variable is a sum of incoming and outgoing tourists. For three countries in the index (Croatia, Czech Republic and Greece) data for outgoing tourists are missing. In these cases (exceptionally) an average of data of all countries on outgoing tourists was added to available country data on incoming students to obtain the sum. 6

9 The variables are presented in the same conceptual structure that emerged from the globalisation definition in the previous section, namely divided into three spheres: an economic, a political and a social sphere. It is important to mention already at this point that these three spheres (economical, political and social) are not to be confused with subdimensions of the index that will later be constructed via a statistical approach (principal component analysis). To make sure that the important aspects of globalisation are accounted for, however, a structured approach based on theoretical reasoning is necessary when selecting individual indicators. This means that each of the three defined spheres should be represented by at least a couple of indicators. Additionally, such a breakdown of the globalisation process into spheres enables a more pragmatic variable search. Of the 21 indicators used in the New Globalisation Index (NGI), 5 do did not appear in the globalisation indices published until now 2. As far as the variables that already appeared in other indices are concerned, the reasoning for including them here is similar to that provided by other authors and some of these variables are becoming quite established as indicators for certain dimensions of globalisation through repeated use. Consequently, they do not require lengthy introduction and will be presented only in a short manner here. Maybe more interesting will be to present the 5 variables being newly introduced here as possible variables for globalisation indices. 2 This refers to the following 6 globalisation indices: G-Index by World Markets Research Centre (Randolph, 2001), ATK/FP Globalization Index by A.T. Kearney and Foreign Policy Magazine (A.T. Kearney/ Foreign Policy, 2007), KOF Index of Globalization by ETH Zurich (Dreher, 2006), GlobalIndex by TransEurope research program of the European Science Foundation (Raab, et al., 2008), CSGR Globalisation Index by the Centre for Study of Globalisation and Regionalisation at Warwick University (Lockwood & Redoano, 2005) and Maastricht Globalisation Index (MGI) by International Centre for Integrated assessment and Sustainable development (ICIS) at Universiteit Maastricht (Martens & Raza, 2008). 7

10 In Table 1, all variables that are new or changed as compared to previously published globalisation indices are marked in bold. Several innovations can be found in the economic sphere. Firstly, the trade in goods is separated from trade in services and treated specifically in order to introduce the geographical distance factor to the index (more details on this in the next section). Secondly, a stock variable has been added next to the flow value for portfolio investment, which is especially important for very volatile variables such as this one. Additionally to trade and finance, two completely new variables have been added to the economic sphere, namely patent and trademarks applications by non-residents. Introducing variables from the intellectual property and innovation domain seemed to add a valuable notion of possibilities, dominance and dynamics of introduction of new products and R&D activities by foreign companies to the domestic market. In the political sphere, three out of four variables are very common in most globalisation indices, clearly representing country s political relations and involvement with the rest of the world. A new variable counts the number of ratified international environmental agreements and in this way introduces environmental issues to the index. In the social sphere, a new variable are the outgoing students 3, concentrating on mobility of young highly educated population, leaving home country only for education purposes or academic exchange. In some globalisation indices, number of internet users is taken as a proxy for the internet variable. This seems like a loose match, as it does not have an explicit international component. International internet bandwidth seems to better describe what is intended to be measured. 3 Data for Germany are unavailable for this variable (outgoing students), so an average of data for France, Switzerland and Austria was taken as a proxy. The idea behind the selection of countries was certain similarity a big developed European country (France) and culturally similar countries with German-speaking population. 8

11 4. The Effect of Geographical Distance Coming back to the definition of globalisation after the indicators for the main spheres have been collected, the geographical distance factor is left to be accounted for. As the definition clearly states, globalisation involves interaction of actors across large distances. Otherwise, we run the risk of confusing regionalisation for globalisation. To give an example, more than 70% of Austria s foreign trade is restricted to the EU area, with about 30% of its exports and 40% of its imports flowing to and from its main trading partner - Germany. To mark this high degree of interconnectedness of EU members as globalisation would be erroneous, as by measuring in this way we may end up measuring regional integration rather than globalisation. This can be avoided through a simple weighting of the bilateral relationships with geographical distance, which will be shown here on the example of the trade in goods variable. The trade in goods variable in this case is a sum of bilateral trade volumes multiplied by the geographical distance between respective countries. The described weighting procedure favours countries that trade most with distant partners, literally across the globe, while trade with neighbouring countries is less emphasized. Austria s trade with China says more about country s globalisation level than the trade with its neighbouring countries does and should therefore receive a higher weight in the indicator. Which countries move up or down in ranking of the trade in goods variable after introducing distances as described above can be seen in Table 2. 9

12 Table 2: Difference in rankings for 'trade in goods' variable with and without distanceweights COUNTRY Simple Distanceweighteweighted Difference COUNTRY Simple Distance- Difference Malaysia Chile Belgium Canada Slovak Rep Mexico Estonia Morocco Czech Rep Portugal Hungary Bolivia Bulgaria Venezuela Slovenia Norway Belarus Turkey Lithuania Georgia Malta Indonesia Moldova Iceland Netherlands Armenia Philippines Russia Ireland Uruguay Latvia Spain Tunisia Italy Ukraine New Zealand Kazakhstan France Austria South Africa Mauritius El Salvador Kyrgyz Rep Cyprus Croatia UK Switzerland Burundi Korea, Rep Argentina Romania Peru Honduras Panama Israel Banglagesh Sweden Colombia Azerbaijan India Finland Australia Germany Greece Poland Japan China Brazil Denmark USA As expected, almost all European countries moved downwards from their positions after distances have been accounted for. Germany and Spain lost 10 positions, France 15. Denmark moved 23 positions backwards in rank, Austria even 34, from position 20 to 54, which sets it behind all the BRICSAM 4 countries (Brazil, Russia, India, China, South Africa and Mexico). United States, on the contrary, gains 14 positions in rank after the change. The weighting procedure also favours Latin American countries. Brazil, Chile, Uruguay, Peru and Argentina 4 BRICSAM term denoting the 6 large emerging economies (Manmohan, 2008). 10

13 all move by more than 30 positions towards higher rankings. Main trading partners of developing countries generally tend to be the United States and certain European countries rather than their immediate neighbours, which may explain these movements in ranks. Distances used as weights come from CEPII 5 and they are weighted so that they take geographic distribution of population inside each nation into account. A small drawback of including the geographical distances in kilometres in the measure might be the size of countries. With a simple geographical measure, a large country turns out to be farther away from its neighbours than a small country. For instance, the measured distance between USA and Mexico is larger than the one between Austria and Slovenia. Nevertheless, this should not be seen as a problem in the example of trade. It is a commonly discussed issue that classical openness measures favour small countries over large ones, as they tend to be more open to foreign trade precisely because of being small and therefore more dependent on foreign resources and markets. With this variable, however, that automatically puts more weight on trade of large countries, this problem becomes neutralized instantly. More details on the influence of country size on the variables of the index are presented in the next section. The described adjustment demonstrated that the geographical distant weights are important for carving out the special feature of globalisation as opposed to regional integration. Of course, it would be desirable to treat other variables in this way to. Due to a lack of bilateral data for most variables, this is difficult to achieve at the moment. However, the distance parameter could be used in for some other economic variables (FDI, portfolio investment) 6 if the index would have been limited e.g. to OECD countries. In any case, further research and 5 Centre d'etudes Prospectives et d'informations Internationales, available at (last retreived March 29, 2009) 6 Some bilateral data for financial variables can be found in UNCTAD, OECD and IMF databases. 11

14 academic discussion in the field of globalisation measurement may induce some new incentives for database construction in the researcher community. 5. Other Adjustments of the Data To become suitable for use in a composite index, the variables have to go through some additional adjustments. As databases are mostly not perfectly filled with data, some imputation of missing data needs to be performed. For values missing in the middle of the sample, linear interpolation was used for this purpose. For values missing at the end of the sample, the last available value was copied. Once the data set is complete, the standard steps include dealing with outliers and normalisation of data. Through a process called winsorisation, extreme values at the ends of the sample have been trimmed to 2.5 and 97.5 percentile values, so as to leave out the possible outliers (Esty et al., 2005). Afterwards, different variables have to be scaled to become comparable through a process of normalisation. Normalisation procedure used here is quite simple and was done using the Min-Max equation 7 on annual basis, whereby all variables take values between 0 and 1. The last step before the grouping of variables into subindices and their aggregation concerns the influence of country size on globalisation outcomes and how this problem can be treated. A common critique on the first globalisation indices (ATK/FP, KOF) was the fact that they favour smaller countries, which were regularly top-ranked in any globalisation index. The case for correcting for country size in the globalisation indices sourced in arguments concerning trade openness, following the argument that small countries have more incentive 7 Using the following formula: ; with x ct = value of the variable X for country c in the year t; max(x t ) and min(x t ) denoting minimal and maximal value for the year t. Using this equation, every value is substituted with a new indicator (I c ) for country c, adopting a value between 0 and 1. 12

15 to trade internationally because they are more dependent on foreign resources and markets (Brahmbhatt, 1998). Lockwood (2004) suggested a correction for that small country bias in his critique on the ATK/FP Globalization Index. Correcting the indicators for size is done by regressing the relevant globalisation variables of the index on measures of country size. Residuals from such estimation should provide a better measure of underlying policies for promoting globalisation than the unadjusted variables themselves. Country size is thereby represented by two different measures, namely the land area and population size. 8 Lockwood (2004) considered such an adjustment to be relevant only for economic variables. This argument does not seem to be justified, as it is quite feasible to apply the same argumentation for other globalisation variables too. It seems to bee reasonable to presume that bigger countries will, for example, have more foreign embassies than small countries, which often find themselves grouped together, having one foreign diplomacy office in charge of a whole region. This is why the adjustment procedure will be applied to all the variables in the index, to check for possible significant effects of country size. As the chosen regressors, population and country size, display high correlation, they turn out to be insignificant when measured at the same time. Nevertheless, for all variables where the regressions with these control variables proved any significance at all, it was clear that one of these variables is considerably more significant than the other. So, the globalisation index variables that proved to be affected by area size were regressed only with this variable, and vice versa for the ones significantly affected by population size. Table 3: Regression results of the adjustment for country size AREA coefficient p-value POPULATION coefficient p-value trade in services -0,4710 0,0025 trademarks -0,6254 0,0030 tourism -0,3978 0,0365 books -0,5129 0,0105 transfers -0,4954 0,0011 newspapers -0,4131 0,0362 int. organisations 0,4126 0,0353 embassies 0,7125 0, A dummy variable for a country being landlocked or having a coast has also been tested, as it has been argued that this structural characteristic has a similar effect on the openness of a country (Lockwood, 2004; Martens & Raza, 2008; Brahmbhatt, 1998). Our calculations, however, showed no significance for this measure. 13

16 Table 3 lists all the variables that were significantly affected by the two control variables representing country size. As expected, trade openness is affected negatively by country size. This suggests that larger countries should move up in the openness ranking after the correction. It may seem confusing that only trade in services was found significantly affected. There is a good explanation for it. The trade in goods variable, namely, showed no significance in the regression after the adjustment for geographical distance took care of the small country bias by putting more weight on trade openness of large countries (as discussed earlier). 9 The political variables (international organisation participation and foreign embassies in country) also fulfilled the expectations and correlated positively to country size. For the eight variables in Table 3, residuals will be used instead of the unadjusted values in the aggregation to the globalisation index. For all the other variables, the effect of country size proved to be insignificant. 6. Construction of Factors and Weights For constructing the globalisation dimensions and variable weights, the principal component analysis (PCA) was used. Dimenstions and weights constructed in such manner reflect the statistical characteristics of the dataset. In this way, the PCA enables helpful insights by testing to what extent these statistical characteristics back up the theoretical framework. When juggling as many variables as there are in composite indicators of complex social phenomena, recognizing the structure of the relationships between variables is not straightforward, due to the complexity and multidimensionality of the set. To understand these relationships would nevertheless be very important for better grasping the phenomenon 9 The unadjusted trade in goods variable (i.e. without distance-related weights) proves significant correlation to country size area (coefficient: , p-value: ). 14

17 being measured. With help of the PCA, it is possible to simultaneously analyse multiple variables in order to discover the patterns of their interrelationships and explain them in terms of their common components. The first step will be to group the variables into 3 factors with help of the PCA and check if the statistical result coincides with the preconstructed theoretical spheres (economic, political and social) from the globalisation definition that served as a guidance for the variable search. In the second step, the three dimensions and individual variables will be assigned weights with which they will enter the overall index. For a detailed and elaborate introduction to PCA and other methods of multivariate analysis, see Hair et al. (2006). PCA creates new factors or components in form of linear combinations of the original data, in a way that the first component absorbs as much as possible of variance of the whole dataset. Subsequently, the second component accounts for the maximum possible proportion of the remaining variance, and so on (JRC/OECD, 2008). This leaves the individual factors uncorrelated (orthogonal), while maximizing the variance absorbed by the principal components. After extracting and rotating the first 3 components, the individual variables tend to load on one particular component more significantly, the rest of their variance being scattered among the other factors. Hence, a certain grouping of the existing variables between the three components takes place. As the individual components are independent of each other, it is possible to say that the variables grouped in different factors really do measure different independent dimensions of globalisation. Ideally, these would coincide with the theoretical dimensions that were defined in the first chapter. In each case, it is a good way to check by means of a statistical method if the selected variables do in fact represent the dimensions they are supposed to stand for. Moreover, a favourable structure of the groups would even allow labelling the individual components basing on conceptual interpretation of 15

18 factor loadings. As can be seen from Table 4, the emerging factor solutions really do resemble the theoretical assumptions substantially. Table 4: Dimensions and Weights FINANCE 37% TRADE AND POLITICS 32% SOCIAL 31% FDI stock 19% trade in goods 11% trade in services 14% FDI flow 13% trademarks 14% migration 9% portfolio stock 20% patents 14% tourism 15% portfolio flow 14% transfers 10% telephone 15% income payments 19% env. agreements 16% books 11% internet 15% int. organisations 18% newspapers 14% embassies 16% outgoing students 12% peacekeeping 11% This is most obvious for variables supposed to represent the social dimension (books and newspapers, migration, tourists, etc.). The third component (factor) is constituted almost exclusively by this group of variables, which is why it can remain with a social label. Political variables (embassies, environmental agreements, etc.) are also mostly grouped within the same component (factor 2), exception being the peacekeeping missions which ended up in the social dimension. The economic dimension proved to be more diverse than assumed in the theoretical introduction. Even so, a consistent conceptual interpretation for this can be found, as the economic dimension obviously got divided into a financial part and a trade-and-business part. All the financial variables are strongly correlated with the first component, while trade in goods and trademark and patent indicators moved to join the political variable in the second component. This is why the second factor was labelled with two keywords: trade and politics. The first variable group clearly represents a financial dimension. The fact that the internet variable also loaded to this financial factor has its interpretation too. The large amount of electronic data sent by banks internationally obviously accounts for the largest part of the variable, making it an inappropriate social indicator. Same goes for the international transfers, which may incline towards political variables, accounting presumably to a large part for the governmental transfers. Not 16

19 everything is as imagined in an ideal case. Much rather would one want to see the trade in services variable in the trade/politics dimension than in the social one. This formation may be connected to the correlation of services and tourism variables. This especially applies for small tourism-oriented countries, where income from tourism constitutes a significant part of their GDP. Nonetheless, the overall results of the analysis can be classified as satisfying, having been able to produce conceptually more or less consistent groups of variables according to their statistical characteristics, which can enter the composite index as separate dimensions. Weights ascribed to each of the dimensions reflect the portion of variance explained by each of these factors and are presented in the title row of the Table 4. The weights of individual indicators within the dimensions emerged from a procedure based on PCA analysis of each dimension (Nicoletti et al. 2000). PCA-generated weights are also popular in other published globalization indices (Dreher, 2006; Lockwood & Redoano, 2005; Raab, et al., 2008). 7. Main Results This chapter presents the results of the New Globalisation Index (NGI). The NGI ranking for the year 2005 is presented in Table 5. The results show that Ireland is the most and Belarus as the least globalized of the 70 ranked countries. Later in text, the description and analysis of the results will concentrate on 4 groups of countries that are of interest small developed countries, large developed countries, new EU-member states and the large emerging countries. 17

20 Table 5: New Globalisation Index 2005 rankings and scores Rank Country Score Rank Country Score 1 Ireland 66,45 36 Bulgaria 23,91 2 Switzerland 60,85 Aaa a aaaa 37 Tunisia 23,65 3 Netherlands 58,47 38 Poland 23,60 4 Belgium 56,78 39 Morocco 23,06 5 Malta 52,99 40 Slovenia 22,60 6 Cyprus 47,96 41 Greece 22,49 7 Iceland 46,99 42 China 22,14 8 United Kingdom 46,91 43 Argentina 22,09 9 Austria 45,90 44 Philippines 21,95 10 Sweden 44,74 45 Bolivia 21,81 11 Denmark 43,43 46 Russian Federation 21,80 12 Canada 42,00 47 Latvia 21,36 13 Norway 37,67 48 El Salvador 21,16 14 Estonia 36,06 49 Azerbaijan 21,11 15 France 35,59 50 Venezuela 20,41 16 Slovak Republic 35,43 51 Peru 20,35 17 Germany 34,62 52 India 20,29 18 Finland 34,17 53 Ukraine 20,14 19 Panama 33,76 54 Mexico 20,07 20 Malaysia 33,09 55 Moldova 19,72 21 New Zealand 31,22 56 Colombia 19,41 22 Spain 30,73 57 Indonesia 18,02 23 Australia 30,69 58 Korea, Rep. 17,79 24 Croatia 29,91 59 Japan 17,71 25 Israel 29,15 60 Lithuania 17,61 26 Portugal 28,61 61 Burundi 16,70 27 Italy 28,50 62 Kazakhstan 15,81 28 Czech Republic 28,46 63 Brazil 15,68 29 Hungary 28,06 64 Romania 15,16 30 Chile 27,96 65 Kyrgyz Republic 14,62 31 Mauritius 27,77 66 Banglagesh 13,72 32 United States 26,96 67 Turkey 13,33 33 Honduras 25,47 68 Georgia 12,51 34 Uruguay 25,44 69 Armenia 12,02 35 South Africa 24,46 70 Belarus 10,49 Except for the overall score, a good look into the composition of the rankings is necessary to draw any meaningful conclusions from the results. This composition varies significantly across different countries, which is nicely shown for some example countries in Figure 1. The figure shows to what extent each of the three dimensions (financial, trade-and-politics and 18

21 social) contributes to the final result of a particular country. Eight countries are shown in the graphic, with two countries representing one of the 4 groups. Netherlands and Ireland are examples of small European countries placed in the top 10 of the ranking. It is easy to note that there exists a certain difference in the composition of the globalisation index between them and the large developed countries, such as Germany and United States. New EU members are represented by Slovenia and Slovak Republic in the graphic, while China and India are logical examples for the large emerging countries. Figure 1 already provides the reader with some interesting patterns, which will be further discussed when backed up by numerical results in the following analysis of the results. Figure 1: Contribution of individual dimensions to total globalisation score for selected countries 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% social tradepol finance Top 10: Small European Countries To start with the top 10, the first group to look at are the small European countries, which constitute a lion s share of this group, United Kingdom being the only larger country in the lot. 19

22 Country Table 6: Top 10 (rankings, 2005) total globalisation finance globalisation tradepol globalisation social globalisation Ireland Switzerland Netherlands Belgium Malta Cyprus Iceland United Kingdom Austria Sweden Average A first issue to discuss is the fact that small countries predominantly take up the highest positions in the index. Such a clear predominance of smaller countries was not expected after a correction for country size has been conducted (see section 4). A closer look at the results of the size adjustment (e.g. for the trade in services variable) shows, namely, that countries with the most dramatic, double digit upward surge are all very large countries (Russia, Canada, United States, China, Brazil and Australia), which is a satisfying result. Small countries, however, are not very much affected by the adjustment and therefore do not fall back much in their score results 10. In this way, even if the adjustment still for pushing the bigger countries upward, small countries kept their high scores. However, it is not only size bias that keeps the small countries in their high positions in the index. First of all, the high top 10 places of the index seem to owe much to a high ranking of the financial indicators (see also Figure 1), which did not show significant correlation with country size. Secondly, these small top ranked countries are all European. A question can be posed on whether these small countries are in fact globalized or only very well integrated in their region. A very important thought that arises from the strong European presence in the 10 It is similar with the political variables (e.g. foreign embassies), where very large countries, as expected, experience an significant reduction in their scores, while the scores of small countries change only modestly. 20

23 high positions 11 is a possible confirmation of the critique from the first chapter that globalisation measures do not distinguish between globalisation and regionalisation. European countries are indeed open to international flows of goods, people and ideas. The question is, however, how many of these flows involve cross-continental interactions? A high regional integration level in Europe presumably pushes these countries forward in globalisation indices, without accounting for this important difference. One proposed step into the right direction in this thesis was the introduction of geographical distances in the trade in goods variable. It showed that European countries tend to fall behind once the distances have been accounted for. Only one variable is however not enough to correct for the bias of the European integration processes in the overall index. To go back to the Table 6, representing the top 10 globalized countries, let us look at the structure of this group more closely. The traditionally developed countries of the group (Switzerland, Netherlands, Belgium, United Kingdom, Austria and Sweden) perform quite well (top 20 rank) in all the individual dimensions. It is different for the other four countries (Ireland, Malta, Cyprus and Iceland), which seem to be pushed to the top owing to a very high ranking in the finance or social dimension, while having a much lower score in the trade and politics dimension. Figure 1 already showed that the globalisation of this group of countries is induced by the financial factor to a larger extent than of the other countries represented in the graphic. This is most notorious maybe for Ireland and Iceland, known for their intense integration with the international financial flows. 12 Malta and Cyprus are also interesting. First account of their high ranking may give the impression that the islands were favoured by the geographical distance weighting of the trade in goods variable. However, the 11 Top 15 is, with the exception of Canada, purely European (see Table 5). 12 From the perspective today in 2009, it would be interesting to see the results for Cyprus for the years of the current financial crisis, which may happen to pull it down in this type of ranking. 21

24 drivers of their high result seem to be hiding in the social variable, which happens to be a distinguishing characteristic for many new EU members and will be discussed further below. Big developed countries Even after agreeing that small and finance driven countries own the top 10 ranks, it still may strike one as surprising that some developed countries such as Italy, USA and Japan do not make it even to the top 25. Table 7: Big developed countries (rankings, 2005) Country total finance tradepol social globalisation globalisation globalisation globalisation United Kingdom Canada France Germany New Zealand Spain Australia Italy United States Japan Average A look at the overview of ten big developed countries in Table 7 shows that the lowest scores for these countries lie in the social dimension. This is especially true for Japan, which suggests a rather introverted society and lets Japan land on the last place for the social dimension. It will prove to be just the opposite for the new EU countries in the next paragraph. Generally, the countries in Table 7 tend to have a higher score in the trade and politics domain than in the other two dimensions. Individually, this is true for all the countries of the group but Australia and Spain. Countries of this group that enter the top 20 show very high rankings in that particular dimension. 22

25 New Europe Table 8 gives an overview of the globalisation index composition for the new EU countries, inclusive the two candidate countries, Croatia and Turkey. A notable feature seen in Table 8 is that all countries except for Poland and Turkey rank much lower in the trade and politics dimension than in the other two. So, trade with distant countries, patent applications by foreigners, environmental issues and diplomatic relations seem to be the weaker points for the group. Many of the countries record their highest performance in the social dimension. A look at the individual indicators reveals that these countries have very high scores for imported and exported books and newspapers on one hand, and services and tourism on the other. The latter is especially conclusive, as countries in which tourism accounts for a significant part of their GDP (such as Croatia) lure foreign tourists and sell services to international visitors at the same time. Table 8: "New Europe" (rankings, 2005) Country total finance tradepol social globalisation globalisation globalisation globalisation Malta Cyprus Estonia Slovak Republic Croatia Czech Republic Hungary Bulgaria Poland Slovenia Latvia Lithuania Romania Turkey Average An interesting note may be the large gap between Estonia (overall rank 14, with very high scores in the finance and social dimension) and the other two Baltic countries, Latvia and Lithuania, who dwell in ranks 47 and 60 respectively. This peculiarity was already noticed 23

26 in the first globalisation index (Randolph, 2001), where Estonia left behind its Baltic neighbours. In the discussion of the results, Estonia is referred to as one of the fastest recent globalisers in the world (Randolph, 2001, p. 8). That Estonia has indeed rapidly and successfully embraced trade and foreign investment from the West (Randolph, 2001, p. 8) is also visible in the New Globalisation Index (NGI), where all its trade and finance related indicators have higher scores than in Latvia and Lithuania. It is however not only a pure economically-based phenomenon that surrounds Estonia. Even though his G-Index measured exclusively the economic and technical indicators, Randalph already suggested in his comments that a full explanation as to why Estonia, of the three Baltic states, has led the way and is now included amongst the first wave of European Union (EU) hopefuls must also refer to the Estonians themselves. Estonians are arguably the most socially cohesive, skilled, educated and economically adaptive people of all the former Soviet Republics. (Randolph, 2001, p. 8). These insinuations about the open nature of Estonian people seem to be confirmed in the New Globalisation Index too. Almost all social indicators (including tourism, international trade in books and newspapers, outgoing students, international telephone calls) are significantly higher than in the other two Baltic countries. Emerging giants The last group to be observed here is a very interesting one in globalisation terms, as some of these countries seem to represent the very idea of globalisation to many people. BRICSAM as a term stands for the 6 big emerging countries (Brazil, Russia, India, China, South Africa and Mexico) presented in the Table 9 (Manmohan, 2008). Interestingly, they are all settled in the middle-lower part of the index (with the exception of the quite low ranking of Brazil) and hence less dispersed than the new EU members, even though this group is geographically much more heterogeneous. Latin American emerging countries perform lower than the Asian counterparts, while South Africa ranks as the most globalized of the group, mostly due to a 24

27 higher score in trade and politics. (a very high distance-weighted trade indicator may be reflecting its connectedness to the trading partners in Europe). A general feature of the group is very low score across the finance dimension. Table 9: BRICSAM countries (rankings, 2005) Country total finance tradepol social globalisation globalisation globalisation globalisation South Africa China Russian Federation India Mexico Brazil Average Changes over Time The New Globalisation Index (NGI) was calculated for the period of 1995 to The annual normalisation procedure that was applied to the data is not ideal for time comparisons in a sense that it can not trace changes in score over time 13. Score values from year to year are not comparable, especially if the changes are small. Therefore, statements about how big the progress of globalisation over the years is are not possible. On the positive side, however, the globalisation index contains the same number of countries over the whole observed time period, which increases comparability of rankings significantly, as rankings are very sensitive to a change in number of observed cases 14. A first note on this subject is that the rankings generally do not change much in short term. This is not surprising, as globalisation is a long-winding process in which sudden changes are normally not expected. Less than 20 countries change their ranking by more than 10 places 13 The annual normalisation provides a better snapshot of globalization ranks for an individual year, which is why it was chosen. For pros and cons of annual normalisation as opposed to the panel normalisation, see Lockwood (2004). 14 Globalization Indicators that trace globalization over long time periods do not keep the same number of countries for each year, which makes the time comparisons less conclusive (e.g. KOF, CSGR) 25

28 over the 1995 to 2005 period. Developments of globalisation over time are not easy to follow in such a short period. Real trends are probably visible only over more decades. Due to the annual normalisation that was used here, small changes are hard to spot, as some volatility in rankings from year to year is generated through the normalisation process, which makes the data less comparable. However, strong trends are still well reflected in the results and for some countries the movement direction is quite clear in the observed period. Comparing Croatia, as a candidate country for the EU, with the other candidate (Turkey) and the two newest Union members, provides some interesting insights (see Figure 2). Figure 2: Globalisation over time Bulgaria, Romania, Croatia and Turkey Bulgaria Croatia Romania Turkey Three of the four countries (Croatia, Bulgaria and Turkey) started in similar positions in year 1995 (around position number 50), with Romania lower, among the last 10 countries. Over time, Croatia and Bulgaria climbed to much higher positions, opening a big gap between them and the other two countries. While Romania kept around the same rank throughout the whole period, Turkey even lost more than 10 positions. 15 It is interesting to observe the very 15 This does not have to mean that the globalization process for Turkey is going backwards, as rankings represent relative values. Rather, it can suggest that while stagnating or only progressing slowly in terms of globalization, it was outpaced by other countries that were previously residing in lower positions. Nevertheless, the point of downturn seems to coincide with the stock market crisis in Turkey